In silicon chip cooling, the heat generated by the chip is ultimately removed by a heat sink, which is usually cooled by air or liquid. It is crucial to have a path of high thermal conductivity between the chip and the heat sink. The simplest form of chip cooling consists in a heat sink made of a high conductivity material, ideally copper, directly bonded to the chip. However, direct bonding of a copper heat sink is limited because of mechanical reliability issues. Due to the considerable difference in thermal coefficients of expansion (TCE) of silicon and copper, mechanical stress often leads to delamination or formation of cracks between the silicon chip and the copper heat sink.
Several methods that have been used to address this problem include, for example, interposing a thermal paste between the chip and the heat sink to eliminate stress. A major drawback, however, is the low thermal conductivity of the paste. Another approach is to provide bonding between the copper heat sink and the silicon chip via an intermediate spreader (usually made of SiC) with TCE closely matched to silicon. This approach adds to process complexity, costs, and non-optimized thermal conductivity.
A recent approach involves the use of a liquid metal layer, e.g., gallium, or various alloys that include gallium, between the chip and the heat sink. This is disclosed in U.S. patent application Ser. No. 10/665798, “METHOD AND APPARATUS FOR CHIP COOLING”, filed on Sep. 18, 2003. However, alternative cooling structures with improved thermal matching and mechanical reliability are still needed.